python - Resampling Error : cannot reindex a non-unique index with a method or limit

Question

I am using Pandas to structure and process Data.

I have here a DataFrame with dates as index, Id and bitrate. I want to group my Data by Id and resample, at the same time, timedates which are relative to every Id, and finally keep the bitrate score.

For example, given :

df = pd.DataFrame(
{'Id' : ['CODI126640013.ts', 'CODI126622312.ts'],
'beginning_time':['2016-07-08 02:17:42', '2016-07-08 02:05:35'], 
'end_time' :['2016-07-08 02:17:55', '2016-07-08 02:26:11'],
'bitrate': ['3750000', '3750000'],
'type' : ['vod', 'catchup'],
'unique_id' : ['f2514f6b-ce7e-4e1a-8f6a-3ac5d524be30', 'f2514f6b-ce7e-4e1a-8f6a-3ac5d524bb22']})

which gives :

This is my code to get a unique column for dates with every time the Id and the bitrate :

df = df.drop(['type', 'unique_id'], axis=1)
df.beginning_time = pd.to_datetime(df.beginning_time)
df.end_time = pd.to_datetime(df.end_time)
df = pd.melt(df, id_vars=['Id','bitrate'], value_name='dates').drop('variable', axis=1)
df.set_index('dates', inplace=True)

which gives :

And now, time for Resample ! This is my code :

print (df.groupby('Id').resample('1S').ffill())

And this is the result :

This is exactly what I want to do ! I have 38279 logs with the same columns and I have an error message when I do the same thing. The first part works perfectly, and gives this :

The part (df.groupby('Id').resample('1S').ffill()) gives this error message :

ValueError: cannot reindex a non-unique index with a method or limit

Any ideas ? Thnx !

Answer 1

It seems there is problem with duplicates in columns beginning_time and end_time, I try simulate it:

df = pd.DataFrame(
{'Id' : ['CODI126640013.ts', 'CODI126622312.ts', 'a'],
'beginning_time':['2016-07-08 02:17:42', '2016-07-08 02:17:42', '2016-07-08 02:17:45'], 
'end_time' :['2016-07-08 02:17:42', '2016-07-08 02:17:42', '2016-07-08 02:17:42'],
'bitrate': ['3750000', '3750000', '444'],
'type' : ['vod', 'catchup', 's'],
'unique_id':['f2514f6b-ce7e-4e1a-8f6a-3ac5d524be30', 'f2514f6b-ce7e-4e1a-8f6a-3ac5d524bb22','w']})

print (df)  
                 Id       beginning_time  bitrate             end_time  
0  CODI126640013.ts  2016-07-08 02:17:42  3750000  2016-07-08 02:17:42   
1  CODI126622312.ts  2016-07-08 02:17:42  3750000  2016-07-08 02:17:42   
2                 a  2016-07-08 02:17:45      444  2016-07-08 02:17:42   

      type                             unique_id  
0      vod  f2514f6b-ce7e-4e1a-8f6a-3ac5d524be30  
1  catchup  f2514f6b-ce7e-4e1a-8f6a-3ac5d524bb22  
2        s                                     w  

df = df.drop(['type', 'unique_id'], axis=1)
df.beginning_time = pd.to_datetime(df.beginning_time)
df.end_time = pd.to_datetime(df.end_time)
df = pd.melt(df, id_vars=['Id','bitrate'], value_name='dates').drop('variable', axis=1)
df.set_index('dates', inplace=True)


print (df)  
                                   Id  bitrate
dates                                         
2016-07-08 02:17:42  CODI126640013.ts  3750000
2016-07-08 02:17:42  CODI126622312.ts  3750000
2016-07-08 02:17:45                 a      444
2016-07-08 02:17:42  CODI126640013.ts  3750000
2016-07-08 02:17:42  CODI126622312.ts  3750000
2016-07-08 02:17:42                 a      444

print (df.groupby('Id').resample('1S').ffill())

ValueError: cannot reindex a non-unique index with a method or limit

One possible solution is add drop_duplicates and use old way for resample with groupby:

df = df.drop(['type', 'unique_id'], axis=1)
df.beginning_time = pd.to_datetime(df.beginning_time)
df.end_time = pd.to_datetime(df.end_time)
df = pd.melt(df, id_vars=['Id','bitrate'], value_name='dates').drop('variable', axis=1)

print (df.groupby('Id').apply(lambda x : x.drop_duplicates('dates')
                                          .set_index('dates')
                                          .resample('1S')
                                          .ffill()))

                                                    Id  bitrate
Id               dates                                         
CODI126622312.ts 2016-07-08 02:17:42  CODI126622312.ts  3750000
CODI126640013.ts 2016-07-08 02:17:42  CODI126640013.ts  3750000
a                2016-07-08 02:17:41                 a      444
                 2016-07-08 02:17:42                 a      444
                 2016-07-08 02:17:43                 a      444
                 2016-07-08 02:17:44                 a      444
                 2016-07-08 02:17:45                 a      444

You can also check duplicates by boolean indexing:

print (df[df.beginning_time == df.end_time])
2        s                                     w  
                 Id       beginning_time  bitrate             end_time  
0  CODI126640013.ts  2016-07-08 02:17:42  3750000  2016-07-08 02:17:42   
1  CODI126622312.ts  2016-07-08 02:17:42  3750000  2016-07-08 02:17:42   

      type                             unique_id  
0      vod  f2514f6b-ce7e-4e1a-8f6a-3ac5d524be30  
1  catchup  f2514f6b-ce7e-4e1a-8f6a-3ac5d524bb22